The AI-integrated traffic surveillance cameras will be used like electronic patrol officers, the department said.

The department recently launched its new Command Information Center, equipped with advanced technologies and AI applications. 

Colonel Pham Quang Huy, Deputy Director of the department, said the command center will fully meet all the requirements for state management in the traffic sector, including data on vehicle registration, inspection, taxes, customs, healthcare, and data from other professional units within the ministry related to people and vehicles.

In terms of scale, the center is a level-1 management center and will, in the future, be connected to 34 smaller centers in provinces and cities, incorporating data from all professional equipment of the entire Traffic Police force.

Data collected from cameras will be transmitted to the center, where the AI system analyzes it to identify behaviors and automatically transfers the data to a management file (folder).

For AI cameras, after detecting a violating vehicle, the system immediately extracts related images/clips along with details such as the route, time of violation, and behavior. Based on vehicle registration data, the system can instantly identify the vehicle owner. 

Currently, AI can detect over 20 types of violations and is continuously being updated.

According to Huy, when a violation is detected, the system records images, license plates, and the location of the violation. Simultaneously, using the license plate, AI automatically cross-references the Traffic Police Department's database to identify the vehicle owner and driver. 

The data is then transferred to the violation processing software without requiring intervention from officers or soldiers.

Evaluating the effectiveness of applying AI to monitoring and detecting violations, Colonel Pham Quang Huy said that in cases when a vehicle's license plate is blurry, AI can cross-reference data and accurately identify the plate with up to 90 percent accuracy.

Notably, through a pilot program installing AI traffic surveillance cameras on Pham Van Bach Street (Cau Giay District, Hanoi), from 00:00 to 12:00 on September 24, authorities detected a total of 1,798 violations.

In the near future, through the violation processing software, via the Traffic Police app (for traffic police) and VNeTraffic app (for the public), vehicle owners will receive notifications within two hours of a violation.

Behaviors such as individuals carrying swords, spears, or other weapons on the road are also identified and flagged by AI to alert on-duty forces. All faces are recorded and can be retrieved in public disorder incidents, making it easier for authorities to track down suspects.

"The AI system also assists the Traffic Police in reviewing and reconstructing a vehicle's route based on characteristics such as paint color and license plate to locate vehicles that flee after causing traffic accidents," Colonel Huy added.

"We are collecting images and data for officers, soldiers, and technicians to train AI in detecting errors, such as drivers or passengers in cars not wearing seat belts. The use of AI-integrated traffic surveillance camera systems can be seen as patrolling in the electronic environment," Colonel Pham Quang Huy stated.

Additionally, the Command Information Center manages the entire Traffic Police force during patrols and controls through a digital map system.

"We can monitor the locations where officers and soldiers are working. At the same time, we can extract specific details such as the officer’s name, badge number, and phone number for contact," he said.

The center can oversee all Traffic Police patrol vehicles, including information on the number of vehicles in operation, those stopped to handle violations, which unit a task force belongs to, how many members it has, what tools they are equipped with, and what specific tasks they are performing on which route. This allows for coordination in case of urgent incidents that need to be resolved quickly.

Another critical task of the Command Information Center is to support the investigation and resolution of traffic accidents.

"For example, a car causes an accident and flees. Based on witness statements, the driver was in a red car, wearing a white shirt, and driving a five-seater vehicle," Colonel Huy posed as a scenario.

Officers at the center can immediately filter how many red cars passed through specific routes during that time frame. Then, based on other identifying features, they analyze which vehicle most closely matches the witness’s description. The AI system also reconstructs the route map of the suspected vehicle to aid in tracking and investigating the accident.

Vu Diep